Re: machine thinking

From: Rick Craik (rick_at_@icebergideas..com)
Date: 06/27/04 

Date: Sun, 27 Jun 2004 14:36:03 -0400

"ray scanlon" <rscanlon@nycap.rr.com> wrote in message
news:363d693e.0406260755.27ce420d@posting.google.com...
> Rick Craik writes:
>
> > The only predicate I was after is the state of synapses. Your previous
> > post was remarkably focused on inhibitor signals. For example, the
> > state of the synapses for a model neuron may have to follow a
> > situational law based on logic. The signals arriving may not be
> > of that form of logic, but the situation could be reduced to one such
> > logic statement. Something like synapses being reduced to OR and
> > XOR relations for positive and negative associations. Although
> > I see it as a bit more complex for a neuron in a group, the basic
> > idea is there.
>
>
> In my opinion, this is a debate over whether the predicate calculus is
> a proper vehicle for describing the thinking brain; I think not. If we
> exclude the "middle", the whole world can be divided into those things
> that are and those things that are not. We can denote those things
> that are with the symbol "+", and those things that are not with the
> symbol "-". All are agreed.

I have to disagree. We can describe things as are true and are not,
or exclusively we describe things as are false and are not. The middle
is not excluded when combined, they are things that are described either
as not true or not false. We denote these things with the
symbols +, - and 0, zero is the middle joining the two notations as
+ and 0, or as - and 0. I think this is applicable to your opinion as for
usage, it makes a fundamental assumption that a neuron does not make.

I have no issue with your opinion of predicate calculus for
describing the thinking brain, I have an issue with you turning
it into a debate about it. What about the non-thinking neuron,
more in-line with your original post in this thread.
I think you are trolling.

>
> But when considering a thinking machine (not the brain, a machine),
> this approach has not worked out too well. The problem is the
> exponential increase in complexity as we increase the number of
> place-holders. The programmers of chess playing machines abandoned
> this approach a long time ago.

In my opinion, if you try to implement a system with an encoding
system that can describe anything, your bound to this problem.
Notably, a given problem has _more_ placeholders in a
system of where each element of the notation is less descriptive.
This would suggest that "exponential increase in complexity" is not
the main issue when fundamental assumptions may be the issue.

>
> You say, "The signals arriving may not be of that form of logic, but
> the situation could be reduced to one such logic statement". This is
> the sort of hand-waving that is beloved of professors. They say,
> "True, in principle". I would also say, "True in principle". Now what!
>
> I say, in principle, "A thinking brain can be described". Then I say,
> "But not with the predicate calculus". The first is a statement of
> hope, the second of experience. Fiddle-Dee-Dee.
>
> I know, from personal experience, that the predicate calculus is
> beautiful. I also believe it is not the way to go. The predicate
> calculus is great on the black board before a philosophy seminar, but
> neuroscientists do not think much of it.
>
> ray

Relevant Pages